Mortality surrogates in combined pulmonary fibrosis and emphysema

An Zhao, Eyjólfur Gudmundsson, Nesrin Moğulkoç, Coline van Moorsel, Tamera J. Corte, Chiara Romei, Robert Chapman, T Wallis, Emma Denneny, Tinne Goos, Recep Savaş, Asia Ahmed, Christopher J. Brereton, Hendrik W. van Es, Helen E. Jo, Annalisa De Liperi, Mark Duncan, Katarina Pontoppidan, Laurens De Sadeleer, Frouke van Beek, Joseph Barnett, Gary Cross, Alex Procter, Marcel Veltkamp, P. Hopkins, Yuben Moodley, Alessandro Taliani, Magali N. Taylor, Stijn E. Verleden, Laura Tavanti, Marie Vermant, Arjun Nair, Iain Stewart, Sam M. Janes, Alexandra L. Young, David Barber, Daniel C. Alexander, Joanna C. Porter, Athol U. Wells, Mark G. Jones, Wim Wuyts, Joseph JacobmedRxiv 2023 · DOI

Abstract

Idiopathic pulmonary fibrosis (IPF) with co-existent emphysema, termed combined pulmonary fibrosis and emphysema (CPFE), is associated with physiological complexities that complicate standard disease monitoring. While Forced Vital Capacity (FVC) decline is the gold-standard surrogate for mortality in IPF, its reliability in CPFE is questionable due to the opposing mechanical effects of fibrosis and emphysema. This study examined two independent cohorts to identify more accurate functional measures of disease progression. The results demonstrate that in patients with >10% emphysema, 1-year decline in diffusing capacity for carbon monoxide (DLco) is a superior predictor of mortality compared to FVC. Furthermore, in non-CPFE IPF patients, a 5% FVC decline threshold was found to be as predictive of mortality as the traditional 10% threshold, suggesting a need for more sensitive monitoring across the IPF spectrum.

Introduction

Idiopathic Pulmonary Fibrosis (IPF) is a progressive and fatal lung disease characterized by scarring of the lung parenchyma. Clinicians typically monitor disease progression using Forced Vital Capacity (FVC)—the total volume of air a patient can forcibly exhale. A 10% annual decline in FVC is widely accepted as a "mortality surrogate," signaling a high risk of death and often serving as a primary endpoint in clinical trials.

However, a significant subset of IPF patients also presents with emphysema, a condition characterized by alveolar destruction and lung hyper-inflation. In these "Combined Pulmonary Fibrosis and Emphysema" (CPFE) patients, the physiological dynamics are altered: the restrictive nature of fibrosis (which reduces lung volume) is countered by the obstructive nature of emphysema (which increases it). This "tug-of-war" can mask disease progression, as FVC may remain deceptively stable even as the patient’s clinical status deteriorates. This study sought to identify more robust mortality surrogates for the CPFE population.

Methodology

The researchers conducted a comprehensive analysis across two large, international patient groups: a Derivation Cohort (n=500) from Turkey, the Netherlands, and Italy, and a Replication Cohort (n=510) from the UK, Belgium, and Australia.

Subgrouping and Scoring

Patients were categorized based on visual emphysema extent on CT imaging. The study tested thresholds of 10% and 15% emphysema to define CPFE. Additionally, the researchers employed an unsupervised machine learning model called SuStaIn (Subtype and Stage Inference). Originally developed for neurodegenerative disease research, SuStaIn was used to identify distinct disease subtypes based on the temporal evolution of fibrosis and emphysema.

Statistical Analysis

Linear mixed-effects (LME) models were used to estimate the annual rate of decline for both FVC and DLco (Diffusing Capacity for Carbon Monoxide), which measures the efficiency of gas exchange. Multivariable Cox regression models then assessed the association between these functional declines and overall mortality.

Key Insights

The Volume Masking Effect

The study confirmed that in patients with significant emphysema, FVC is an unreliable metric. As fibrosis progresses, it can cause traction bronchiolectasis (pulling open airways), which, when combined with emphysematous hyper-inflation, artificially preserves lung volume. Consequently, a patient may be nearing respiratory failure while their FVC remains relatively stable.

DLco as a Superior Signal

Unlike FVC, DLco measures the functional surface area available for gas exchange. Because both fibrosis and emphysema destroy this surface area, DLco decline is not subject to the masking effect seen in FVC. In patients with >10% emphysema, DLco decline provided a clear, consistent signal of disease progression and mortality risk.

Redefining Thresholds for Non-CPFE IPF

For patients without significant emphysema, the study challenged the traditional 10% FVC decline threshold. The data indicated that a 5% FVC decline was equally predictive of mortality. Adopting this more sensitive threshold could identify a significantly larger proportion of at-risk patients much earlier in their disease course.

Results

In both cohorts, CPFE patients with >10% emphysema showed a greater smoking history and lower baseline DLco. In multivariable Cox regression analyses for this group, 1-year DLco decline was a significantly better indicator of mortality than FVC decline. In the replication cohort specifically, FVC decline failed to reach statistical significance as a mortality predictor in CPFE, whereas DLco remained robust.

The SuStaIn machine learning model identified a "Matched CPFE" subtype where fibrosis and emphysema progressed concurrently. In this specific group, FVC decline was a particularly poor surrogate for death, while DLco decline maintained a strong predictive value (C-index ~0.73).

Discussion and Clinical Implications

These findings suggest a necessary shift in how clinicians and researchers monitor IPF. For patients where CT imaging reveals >10% emphysema, DLco should be prioritized over FVC as the primary indicator of progression. Furthermore, the results suggest that current therapeutic trials may be using inappropriate endpoints for CPFE patients, potentially excluding or mischaracterizing the progress of a significant portion of the patient population.

While the study relied on visual CT scoring by radiologists, future work should focus on integrating automated, AI-driven CT analysis to improve scalability and precision. Prospective validation of DLco as a primary endpoint in CPFE-specific clinical trials is also warranted to solidify these findings into standard clinical practice.

Summary updated

Abstract

Abstract Background Idiopathic pulmonary fibrosis (IPF) with co-existent emphysema, termed combined pulmonary fibrosis and emphysema (CPFE) may be associated with reduced FVC decline compared to non-CPFE IPF patients. We examined associations between mortality and functional measures of disease progression in two IPF cohorts. Methods Visual emphysema extent (CPFE:non-CPFE: derivation cohort=317:183; replication cohort=358:152), scored on computed tomography imaging subgrouped CPFE patients using either a) 10%, or b) 15% visual emphysema threshold, or c) an unsupervised machine learning model considering emphysema and ILD extents. Baseline characteristics, 1-year forced vital capacity (FVC) and diffusion capacity for carbon monoxide (DLco) decline (linear mixed effects models), and their associations with mortality (multivariable Cox regression models) were compared across CPFE and non-CPFE subgroups. Results In both IPF cohorts, CPFE patients with >10% emphysema had a greater smoking history and lower baseline DLco compared to CPFE patients with <10% emphysema. Using multivariable Cox regression analyses in patients with >10% emphysema, 1-year DLco decline was a better indicator of mortality than 1-year FVC decline. Results were maintained in patients suitable for therapeutic IPF trials. Results were replicated in the >15% emphysema population and using unsupervised machine learning. Importantly, the unsupervised machine learning approach identified CPFE patients in whom FVC decline did not associate strongly with mortality. In non-CPFE IPF patients, 1-year FVC declines >5% and >10% showed comparable mortality associations. Conclusion When assessing disease progression in IPF, DLco decline should be considered in patients with >10% emphysema and a >5% 1-year FVC decline threshold considered in non-CPFE IPF patients.
BibTeX
@misc{zhao2023mortality,
  author = {An Zhao and Eyjólfur Gudmundsson and Nesrin Moğulkoç and Coline van Moorsel and Tamera J. Corte and Chiara Romei and Robert Chapman and T Wallis and Emma Denneny and Tinne Goos and Recep Savaş and Asia Ahmed and Christopher J. Brereton and Hendrik W. van Es and Helen E. Jo and Annalisa De Liperi and Mark Duncan and Katarina Pontoppidan and Laurens De Sadeleer and Frouke van Beek and Joseph Barnett and Gary Cross and Alex Procter and Marcel Veltkamp and P. Hopkins and Yuben Moodley and Alessandro Taliani and Magali N. Taylor and Stijn E. Verleden and Laura Tavanti and Marie Vermant and Arjun Nair and Iain Stewart and Sam M. Janes and Alexandra L. Young and David Barber and Daniel C. Alexander and Joanna C. Porter and Athol U. Wells and Mark G. Jones and Wim Wuyts and Joseph Jacob},
  title = {{Mortality surrogates in combined pulmonary fibrosis and emphysema}},
  year = {2023},
  doi = {10.1101/2023.05.05.23289330},
  howpublished = {medRxiv},
}
Figure